Vehicle Damage Analysis Using Computer Vision: Survey

Abstract

One of the biggest problems in the transportation industry is vehicle damage. The manual inspection of these damaged automobiles takes a long time. Processing vehicle insurance using pictures of damaged cars is a crucial industry with lots of possibilities for automation. A segmentation method for detecting vehicle damage that is based on machine learning. When submitting insurance claims, using photos taken at the scene of an accident can expedite the process and save time and money while also improving driver convenience. This work examines how much automotive damage is worth. This research focuses on computer vision-based studies from the previous five years. This Convolutional Neural Networks serve as the foundation for the methods utilised. Based on the study, a majority of the work focuses on Mask R-CNN, but there is still potential to increase the performance. This research widens the window for auto insurance and damage detection.